In order to predict and reduce Shinkansen noise, it is important to analyze noise sources and develop a high-precision predicting technique.
Firstly, Shinkansen noise elements were classified into the following categories of sound source: current collecting system noise, vehicle lower part noise, aerodynamic noise from vehicle upper part, and concrete bridge structure noise. A method was developed to calculate the contribution of an individual sound source at the measuring point based on actual data measured with a microphone array. A prediction model of Shinkansen noise was also then established based on the results of sound source contributions.
This model can predict noise levels for all Shinkansen vehicles currently in service by entering conditions relating to track, structure, and train speed (Table 1). Since the model uses an energy-based calculation model, this can estimate not only the maximum A-weighted sound pressure level of a train set pass with time-weighted characteristics S (LpA, Smax) , but also an A-weighted sound exposure level of a train set pass (LAE) and an equivalent continuous A-weighted sound pressure level (LAeq, T ).
Statistical analysis of the correlation between predicted values and measured values showed their average difference to be 0.7 dB with the standard deviation being 1.5 dB. This verifies that the method is sufficiently accurate (Fig. 1).